Norwegian law, for your AI

The source text of Norwegian law, where your AI can read it

Lexar is an MCP service. Connect Claude or another MCP client, and it can search, look up and read current Norwegian laws and central regulations — from the source text, with the citation attached and whether the provision is in force, instead of from memory.

Connect Lexar Read the documentation

What is in the corpus

One pinned snapshot of the open data Lovdata publishes. Lexar gives access to exactly this, and nothing beyond it.

759 current consolidated laws
5 112 current central regulations
988 Norsk Lovtidend division I announcements from the current calendar year
6 859 documents in total

Preparatory works (forarbeider) are held alongside the law: about 29,600 documents from Nasjonalbiblioteket and Stortinget. When the store links preparatory works to a provision, reading it returns ranked pointers to them — by default only the high-precision Stortinget links; the rest, including the Nasjonalbiblioteket half, are available on request. Passages that only quote the law back are hidden by default and counted. When nothing is left to show, it says so, and where it can it names the enacting document. Propositions (Prop. and Ot.prp.) after 2005 are mostly not held and appear as designations only.

Two acts, one name — and which one is in force

Ask for a law by name. When the name fits more than one act, you get every candidate, each marked with whether it is in force — Lexar never picks one silently.

resolve("forvaltningsloven")

→ NL/lov/1967-02-10      in_force: true
  Lov om behandlingsmåten i forvaltningssaker (forvaltningsloven)
→ NL/lov/2025-06-20-81   in_force: false
  Lov om saksbehandlingen i offentlig forvaltning (forvaltningsloven)
  in_force_basis: commencement left to the King; no commencement recorded

Live output, 24 September 2026, abridged. Results follow the served snapshot and can change.

Both acts are called forvaltningsloven. Lovdata records no commencement for the 2025 act, and the answer says so, with the reason taken from Lovdata's own commencement header. The same status comes with every search hit and every provision read. When the source does not say, the status is unknown — never a guess.

Ask in your own words

People do not ask the way statutes are written. Lexar keeps a verified map from everyday words to the statute's own terms, adds the statutory term beside your word — never instead of it — and each hit tells your assistant which words matched in the title, a heading, the text, or only through that map.

search("kan huseieren kaste meg ut av leiligheten")

  "kaste meg ut" → oppsigelse, fraviking     (lay_vocabulary)
  leiligheten    → husrommet                 (lay_vocabulary)

1 NL/lov/1999-03-26-17/§9-5   in_force: true
  Husleieloven › § 9-5. Utleierens oppsigelse av tidsubestemt leieavtale
  matched: lay_vocabulary [kaste, leiligheten]

Live output, 24 September 2026, abridged. Results follow the served snapshot and can change.

What your assistant can do

search — search the legal source text in plain words or statutory terms, ranked for Norwegian, with in-force status and where each word matched on every hit
resolve — turn a law name, a written reference such as “§ 15-7 i arbeidsmiljøloven”, or a legacy document ID (such as LOV-2005-06-17-62) into what it names, with every candidate when it is ambiguous
read · outline — read the exact wording of a provision or a whole document, with its in-force status and its preparatory-works pointers, and list the complete structure of an act or of a preparatory work
connections — follow the cross-references the source itself provides, including ones that do not resolve, and page through every preparatory-works pointer for a provision
terms — browse the words used in the corpus by prefix — vocabulary, not definitions
status — see which snapshot is being served and how many documents it holds
operator_guide — a short field guide your assistant reads once: which tool answers which question, and how to read the results. Results also name the next call when one is plainly next.

What Lexar does not contain

No court decisions.
No local regulations.
No English translations of the law.
No historical consolidated versions — the current text only.
No AI-written commentary or interpretation. The law text and the preparatory works are source text and are never rewritten. One separate, labelled layer exists: keyword labels on a small number of documents, drafted with AI assistance and published only after a recorded review.

Lexar is a retrieval and citation tool. It is not legal advice and it does not tell you what the law means. In-force status is read from Lovdata's commencement header only: true means a recorded commencement, not a check for later repeal.

Source and licence

Inneholder data fra Lovdata under Norsk lisens for offentlige data (NLOD) 2.0. NLOD 2.0

The legal text is open data. Lexar does not own it, and you may obtain and reuse it independently under NLOD 2.0. What Lexar provides is its own work around that text: formatting, indexes, citation resolution, and a hosted service that stays available.

Getting started

In Claude, add a custom connector named Lexar with this address and sign in:

https://lexar-api.publifye.com/mcp

Current plans and prices are in the store. See the store

Questions about Lexar

What people use it for

Can I find the law behind my landlord's notice?

Tenants can ask in everyday Norwegian and retrieve relevant provisions with citations and source wording. Lexar's checked vocabulary map adds statutory terms alongside your own words; it does not decide whether your landlord's action is lawful.

Ask your AI: Finn lovteksten om at huseieren vil kaste meg ut av leiligheten.
What it does: operator_guide → search → read
What you get: Relevant source passages, citations and commencement information to examine or take to an adviser.

Where does the law say when an employee can be dismissed?

Employees and workplace representatives can locate the relevant law by name or a written section reference, then read its exact wording. If several documents match, Lexar returns candidates instead of choosing silently.

Ask your AI: Slå opp § 15-7 i arbeidsmiljøloven og vis meg ordlyden med kilde.
What it does: operator_guide → resolve → read
What you get: The provision's exact wording, citation, source URL and in_force basis.

Can I check the preparatory works behind this section?

Lawyers, advisers and students can follow ranked pointers from a provision into the preparatory works Lexar holds. Coverage is incomplete: propositions after 2005 are mostly not held, and lower-confidence pointers require an explicit request.

Ask your AI: Finn forarbeidene som er koblet til denne bestemmelsen, og les de relevante utdragene.
What it does: operator_guide → resolve → read → connections kind=forarbeider → read
What you get: Available preparatory-work passages and their references, with confidence and coverage limitations visible.

Two laws have the same name. Which text am I looking at?

Public-sector staff and anyone checking a citation can see every matching document with the commencement information supplied by the source. Lexar distinguishes true, false and unknown; true records commencement rather than a complete check for later repeal.

Ask your AI: Slå opp forvaltningsloven og vis alle kandidatene med ikrafttredelsesgrunnlag før du leser.
What it does: operator_guide → resolve → read
What you get: Explicit candidates and source-based commencement information so the correct text can be selected.

Can I follow the other sections a provision refers to?

Researchers can follow cross-references the source itself provides, then read the linked text. Unresolved or ambiguous references remain visible, and a citation is not presented as a legal interpretation.

Ask your AI: Vis henvisningene fra denne bestemmelsen og les de relevante bestemmelsene de peker til.
What it does: operator_guide → resolve → connections → read
What you get: A traceable set of source references and the linked wording where it resolves.

Can my study group read an act in its proper structure?

Students and teachers can inspect a document's outline and read the sections they need, including documents without numbered provisions. Large responses provide continuation cursors so an assistant can retrieve the remaining text from the same snapshot.

Ask your AI: Vis strukturen i denne loven og les de aktuelle delene uten å hoppe over fortsettelsen.
What it does: operator_guide → resolve → outline → read
What you get: A document map and cited source passages, with continuation when a response is split.

All questions